DESIGN OF SERVICE SYSTEM FOR INSURANCE BUSINESS FACING CUSTOMER IMPATIENCE USING QUEUING THEORY

Transcription

1 DESIGN OF SERVICE SYSTEM FOR INSURANCE BUSINESS FACING CUSTOMER IMPATIENCE USING QUEUING THEORY Rakesh Kumar 1, Bhupender Kumar Som 2 1 School of Mathematics, Shri Mata Vaishno Devi University, Katra, J&K (India) 2 ACCMAN Institute of Management, 46A/2, Knowledge Park, 3, Greater Noida (India) rakesh_stat_kuk@yahoo.co.in, bksoam@live.com Abstract; Insurance sector is crowded with a number of insurance players. Based on the quality of service and expected returns, customers may be attracted towards other service providers. As a result the customers may discontinue their policies with a particular firm or do not opt to join the firm at all. This behavior of customers is quite harmful for the growth and sustainability of any insurance firm. Firms employ a number of strategies to retain the impatient (reneged) customers and to stop balking (customers not joining the firm at all due to some reasons) for sustaining their business. In this paper, a stochastic queuing model is proposed to study the retention of reneged customers along with customer balking, when any customer retention mechanism is employed. It is envisaged that if the firms employ some customer retention mechanism, then there are chances that a certain proportion of reneged customers may be retained. The cost-profit analysis of the model is also carried out. The optimum service rate of the system is obtained using the classical optimization techniques. The sensitivity in the optimum service rate is studied by varying the mean arrival rate and the mean reneging rate. The effect of the change in mean reneging rate and the change in mean arrival rate on the total expected cost, total expected revenue, and total expected profit are also studied. Key Words: Optimum service rate, Insurance Business, Reneging, Balking, Total Expected Cost

2 1. Introduction: Due to high competition the retention of impatient customers has become a key issue for any business. Insurance business is not exceptional as well. In thrust of maximizing profit companies try to reach more and more customers. Impatient customers may also make some potential customer balk (customers leave without joining the business (queue)) away by communicating their level of dissatisfaction. So companies facing customer impatience are impacted with another blow in form of customer balking resulting in loss of more customers. To minimize above mentioned customer loss, high level of customer service is required which incurs high cost of service. One way to increase profit is by minimizing the cost. Thus, there is a need to enforce optimum service rate so that the cost of providing service can be minimized, the customers are served more satisfactorily, and the total profit can be increased. Cochran et al. [10] explores the concept of customer impatience and the loss due to this the business faces. Wu et al. [12] focus on an M/M/s queue with multiple vacations such that the server works with different service rates rather than no service during vacation period. They generalize an M/M/1 queue with working vacations. A cost function is formulated to determine the optimal number of servers subject to the stability conditions. Direct search and Newton- Quasi algorithm is applied to find an approximate solution. Same algorithm is applied to investigate the M/M/R machine repair problem with second optional repair by Wang et al. [9]. A cost model is derived to determine the optimal number of repairmen, the optimal values of the first essential repair rate, and the second optional repair rate while maintaining the system availability at a specified level. Ke et al. [8] construct the membership functions of the system characteristics of a heterogeneous-server queuing model with fuzzy customer arrival and service rates. Ching et al. [7] find that when the marginal cost of service capacity is low relatively to the revenue per customer, a unique Nash equilibrium exists, in which all servers choose the same service capacity and the expected waiting times are finite. Tadj et al. [11] use a vacation queuing model and develop a set of quantitative performance measures for a two-parameter time allocation policy. Based on the renewal cycle analysis, they derive an average cost expression and propose a search algorithm to find the optimal time allocation policy that minimizes the average cost. Yue et al. [6] study a two-server Markovian network system with balking and a Bernoulli schedule under a single vacation policy, where servers have different service rates

3 Yue et al. further in [4] present analysis for an M/M/R/N queuing system with balking, reneging and server breakdowns. A cost model is developed to determine the optimum number of servers while the system availability is maintained at a certain level. Wang et al. in [1] develop a cost model for an M/M/R queuing system with finite capacity balking, reneging and server breakdowns. Sensitivity analysis of the model along with numerical results is also performed. Pan [5] studies an M/M/1/N queuing model with variable input rates and points out the appropriate service speeds for the business. Yue et al. [3] present an analysis for an M/M/c/N queuing system with balking, reneging and synchronous vacations of partial servers together formulate a cost model to determine the optimal number of servers on vacation. They perform sensitivity analysis through numerical experiments. Literature survey shows the importance of cost modeling and its solution in way to optimize profit. Borst et al. [2] apply queuing theory for designing the call centers. Recently, Kumar and Sharma [13] study a single server, finite capacity Markovian queuing model with reneging, balking and retention of reneged customers. They derive its steady-state solution. 2. Description of the Model: In this paper, we propose a single server, finite capacity Markovian queuing model with reneging, balking, and retention of reneged customers for any insurance firm facing the problem of customer impatience (reneging and balking) and implementing various customer retention strategies as studied by Kumar and Sharma [13]. The customers arrive in the system according to a Poisson process with mean rate. An arrival to the system represents the sale of one insurance policy. There is a single server and the customers are served in order of their arrival. The service time distribution is exponential with parameter µ. Here, the service of a customer represents the claim processing at the maturity of the policy. The insured customers at any stage before maturity may get impatient due to various reasons like dissatisfaction of service, immediate requirement of money, better opportunities with the other insurers etc. and may withdraw their policies. This phenomenon is analogous to reneging in case of queuing theory. The customers get impatient (reneged) following exponential distribution with parameter. As the customer impatience has highly negative impact on the business of the firms, they employ different customer retention strategies to retain their customers. It is envisaged that if the firms employ - 3 -

4 certain customer retention strategies, then there are chances that a certain proportion of impatient customers may be retained. Thus, an impatient customer may be retained in the system with some probability q (say) and he may not be convinced to stay in the system for his complete service with probability 1 q (= p). The customer retention strategies in insurance business can be the convincing of customers by phone calls, better and reliable service, providing better returns, discounts on premium etc. The arriving customers may balk with probability n/n, where n is the number in system and N is the maximum number allowed in the system (i.e. some customer may not decide to purchase the policy due to dissatisfaction about company policies). The steady-state probabilities of the model are given by: N k1 λ P P ; n N n n k 1 N μ k 1 ξp (1) Also for n=n, we get N N ( k1) PN P0 ; n N N ( k 1) p k 1 (2) Using the normalization condition, N n1 P 1, we get n P0 1 N n n1 k 1 1 N ( k 1) 1 N ( k 1) p The expected system size is: N N N ( k1) Ls n P0 N ( k 1) p n1 k 1 (3) (4) 3. Cost Model: In this section, we present the cost-model and perform optimization. Notations: 1/λ = mean inter-arrival time. 1/µ = mean service time

5 . We define the total expected cost (TEC) of the system as: Where the average reneging rate, Rr, the average retention rate, RR and the average balking rate, are given by; - 5 -

6 Let R be the revenue earned for providing service to each customer per unit time, then RLs is the total earned revenue by providing service to average number of customers in the system. Rλ PN, RRr and RRb are the losses in the revenue of the system due to number of lost customers per unit time, due to reneging of customers and due to balking of customers per unit time respectively. Hence, total expected revenue (TER) of the system is given by: Now, total expected profit (TEP) of the system is defined as: Thus, we have the TEC, TER and TEP functions in terms of various parameters involved. The cost profit analysis of the model is performed numerically by using these functions and the results are discussed accordingly. The optimization of the model is also carried out in order to obtain the optimal service rate and to obtain the optimum values of TEC, TER and TEP. The impact of various customer retention strategies on the total optimal profit is also analyzed. It becomes quiet intractable to proceed analytically for optimum value of service rate (µ*) at which the TEC is minimum. Therefore, we have used MATLAB software to get the optimum values using a computational algorithm. Computational algorithm: Step 1: Define variables Step 2: Write the formula of function TEC in terms of Step 3: Obtain critical values of µ Step 4: Find the value of at which TEC is minimum (let it be *) Step 5: Compute the values of TEC, TER and TEP at * Table: 1.1 Variation in Ls w. r. t. Mean reneging rate () at q=0 and q=0.6 We take λ = 2, μ = 3, N = 4-6 -

7 Ls at q=0 Ls at q= From table 1.1 it is observed that the system size is remains lesser at q=0, as there is no retention strategy is followed for reneged customers as a result more and more customers move out of the system with increasing rate of reneging and that results in a reduction in system size. On the other hand when some retention strategy is followed to retain the reneged customers in terms of premium discounts, better service assurance, etc the system size increases comparatively. System size reduces with increasing rate of reneging though. But it is observed that the firm employing some customer retention strategy for the reneged customers increases their system size that ultimately results in more customers in the system which leads to more profit

8 Figure 1.1 Figure 1.1 gives an insight to table 1.1 and provides a comparative analysis between the system sizes at q=0 (when no reneged customer is retained) and at q=0.6 (When 60% of the customers are retained by employing some customer retention strategy) with changing rate of reneging. Table: 1.2 Variation in TEC, TER and TEP with the change in average reneging rate, when service rate is optimized We take λ = 2, q=0.6, N = 4, C b = 7, C s = 4, C h=3, C L =12, Cr =8, CR =45, and R=100. * TER TEC TEP

9 In table 1.2, optimum service rate is derived for every value of reneging rate, as varying rate of reneging affects the system size, hence constant rate of service is unexpected and unprofitable. Because of complexity of manual classical optimization technique a MATLAB program is constructed for classical optimization technique as per the algorithm mentioned and service rate is optimized for varying mean rate of reneging. The profit thus obtained is maximized. The profit obtained decreases from Rs to ; this is due to the fact that increasing rate of reneging reduces the system size and leaves fewer customers in the system. Figure 1.2 Figure 1.2 gives provides a bird s eye view to table 1.2 s analysis. It shows the movement of total expected return, total expected cost and total expected profit. Total expected cost - 9 -

10 increases with increase in reneging rate due to the fact that retention strategy applied to retain the reneged customers involve some cost mentioned by CR. Table: 1.3 Variation in TEC, TER and TEP with the change in mean arrival rate, when service rate is optimized We take ξ = 0.1, q=0.6, N = 4, C b = 7, C s = 4, C h=3, C L =12, Cr =8, CR =45 and R=100. * TER TEC TEP In table 1.3 optimum service rate is derived for varying values of arrival rate, as varying rate of arrival affects the system size, hence constant rate of service is unexpected and unprofitable. Because of complexity of manual classical optimization technique a MATLAB program is constructed for classical optimization technique as per the algorithm mentioned and service rate is optimized for varying rate of arrival. The profit thus obtained is maximized. The profit obtained decreases from Rs to ; this is due to the fact that increasing rate of arrival increases the cost of service, cost of holding, cost of reneged customers and cost of retention of reneged customers. It can be observed that optimum service rate increases with increase in arrival rate, as system has to provide service to more and more customers

11 4. Conclusions: A queuing theory approach is used to model insurance business problems. A single server finite capacity Markovian queuing model with retention of reneged customers, and balking is proposed for any insurance firm facing the problem of customer impatience (reneging and balking) and implementing different customer retention strategies. The cost profit analysis of the model is performed and the impact of varying mean reneging rate and mean arrival rate on the total expected profit of the system is studied. The optimization of the model is performed in order to minimize the total expected cost of the system with respect to the service rate. The analysis carried out in this paper is very useful to any insurance firm. References: [1] K. H. Wang, Y. C. Chang, Cost analysis of a finite M/M/R queuing system with, balking, reneging, and server breakdowns, Mathematical Models of Operations Research, vol. 56, no. 2, pp , [2] S. Borst, A. Mandelbaum, and M. I. Reiman, Dimensioning large call centers, Operations Research, vol. 52, no. 1, pp , [3] D. Yue, W. Yue and Y. Sun, Performance analysis of an M/M/c/N queuing system with balking, reneging and synchronous vacations of partial servers, The Sixth International Symposium on Operations Research and Its Applications, pp , [4] D. Yue and W. Yue, Block partitioning matrix solution of M/M/R/N queuing system with balking, reneging and server breakdowns, Journal of Industrial Management and Optimization, vol. 5, no. 3, pp , [5] Q. Pan, The research and application of M/M/1/N queuing model with variable input rates, variable service rates and impatient customers, International Journal of Engineering and Natural Sciences, vol. 4, no. 4, pp ,

12 [6] D. Yue and W. Yue, A heterogeneous two server network system with balking and a Bernoulli vacation schedule, Journal of Industrial Management Optimization, vol. 6, no. 3, pp , [7] W. K. Ching, S. M. Choi, and M. Huang, Optimal service capacity in a multiple server queuing system: A game theory approach, Journal of Industrial and Management Optimization, vol. 6, no. 1, pp , [8] J. C. Ke, H. I. Hung, and C. H. Lin, Analysis of a queue system with heterogeneous servers and uncertain patterns, Journal of Industrial and Management Optimization, vol. 6, no. 1, pp , [9] K. H. Wang, C. W. Liao, and T. C. Yen, Cost analysis of the M/M/R machine repair problem with second optional repair: Newton Quasi Method, Journal of Industrial and Management Optimization, vol. 6, no. 1, pp , [10] J. K. Cochran and J. R. Broyeles, Developing Non Linear Queuing Regression to Increase Emergency Department Patient Safety Approximating Reneging with Balking, Computers and Industrial Engineering, vol. 59, no. 3, pp , [11] L. Tadj, Z. G. Zang, and C. Tadj, A queuing analysis of multi purpose facility s operations, Journal of Industrial and Management Optimization, vol. 7, no. 1, pp , [12] C. H. Wu, K. H. Wang, J. C. Ke and J. B. Ke, A heuristic algorithm for the optimization of M/M/S queue with multiple working vacations, Journal of Industrial and Management Optimization, vol. 8, no 1, pp. 1 17, [13] R. Kumar, and S. K. Sharma, An M / M / 1 / N Queuing Model with Retention of reneged customers and Balking, American Journal of Operational Research, vol. 2, no. 1, pp. 1-5,